Enhanced brain tumor detection from brain MRI images using convolutional neural networks

The brain is one of the most important organs of a human body. It controls all activities of the body and hence is referred to as the CPU of human body. Fast, accurate and early diagnosis of brain tumors is essential for better treatment plans that potentially result in larger survival rates. Automatic and non-invasive brain tumor detection methods are highly relevant. Continuous efforts by researchers have resulted in a significant increase in brain tumor detection accuracy, but 100% accuracy on testing/validation data is still challenging to obtain. This paper presents a convolutional neural network model that significantly enhances brain tumor classification accuracy with low complexity. To do so, Kaggle's publicly available dataset titled “Brain_Tumor_Detection_MRI”, consisting of 2891 brain MRI images annotated as “yes” (having tumors) or “no” (without tumors), was used. By leveraging advanced deep learning techniques, we achieved an accuracy of 99.31% on the test dataset. This value shows a significant improvement in classification accuracy, showcasing the potential of convolutional neural networks in medical imaging and diagnostic applications.